Pictures Worth A Million Bytes

By MICHAEL D. LEMONICK

Among the flashy hardware and software on display at last week's First World Supercomputer Exhibition in Santa Clara, Calif., the small Cornell National Supercomputer Facility booth attracted attention out of proportion to its size. There, on a large video screen, more than a thousand stars wheeled around a newly formed black hole, an incredibly dense, bizarre entity with gravity so strong that not even light can escape from it. As nearby stars were sucked in by its gravity, the hole grew. By the time the system stabilized, nearly half its stars were gone. Conventioneers were fascinated.

But not as much as some scientists were. Before their equations were converted into computer images, astrophysicists had predicted that only a tenth as many stars in such a system would be eaten by a black hole. This was no isolated case. Across the nation, in disciplines ranging from geophysics to medicine to entomology, scientists are discovering that computer images can sometimes lead to a better understanding of nature. Borrowing a leaf from Hollywood's special-effects book (and in some cases hiring Hollywood technicians), they are converting their data into video form. Because the human brain is exquisitely adept at picking up visual cues, scientists have begun benefiting from what Robert Langridge of the University of California at San Francisco calls "computer-aided insights." Says Langridge, who uses 3-D graphics to model biological molecules: "Computer graphics gives us a window into what is going on, rather than just a scientific result. It has become an experimental tool."

At the University of Massachusetts in Amherst, mathematicians plot complex equations on a computer-graphics terminal, which translates the numbers and symbols into form and color. Watching a curving, perforated object take form on the screen, the mathematicians gradually become convinced that they have produced a new shape with a jawbreaking name: a complete embedded minimal surface with finite topology. Previously only three such shapes were known to exist; topologists have sought and speculated about a fourth for two centuries, but until this moment it has never been proved to exist. The imagery demonstrates that there are an infinite number of such surfaces.

In West Lafayette, Ind., a Purdue University biologist who until recently was building models of viruses by laboriously fastening together hundreds of brass fittings taps away at a computer keyboard. When he is done, he has created on the screen an image of rhinovirus 14 (one of some 113 varieties responsible for the common cold) that can be turned and viewed in three dimensions. Rhinovirus 14 thus becomes the first animal virus of any kind to have its full portrait drawn.

The growing need for electronic imagery rises from the sheer number- crunching power of computers like those shown in Santa Clara. Says Craig Upson, a graphics specialist who last August left a commercial animation firm to join the staff of the National Center for Supercomputing Applications at the University of Illinois: "You find yourself lost in this maze of data because suddenly you can compute far more than you can comprehend." The route to comprehension, he says, is to turn the numbers into images.

NCSA, one of five regional supercomputer centers established since 1985 by the National Science Foundation, is rapidly emerging as a leader in scientific graphics. Last year, for instance, Artist Donna Cox and Computer Scientist Ray Idaszak helped Caltech Astrophysicist Charles Ross Evans produce a short videotape depicting what in theory would occur in the collision of two neutron stars. To the untrained eye, the colliding stars look more like exotic flowers than a cosmic catastrophe. But the colors all have a quantitative meaning: areas colored red are ten times as dense as yellow ones, and yellow represents 100 times the density of blue. "People who are not involved in these calculations might wonder if we couldn't spend our time better doing science than making movies," Evans says. "What they don't understand is that the movies are necessary to the science."

That is true at the microscopic as well as the telescopic level. Michael Rossmann, who modeled the common-cold virus, became a convert to computer graphics after Purdue acquired its first graphics machine. Compared with a physical model, he says, "the computer is much more versatile. We can zoom in as close as we like; we can look at much more complicated structures. We can display the model on all sides and in different colors." In the old days he would often mark different atoms in his brass models with colored yarn -- which kept falling off. "The old methodology seems so cumbersome now, even laughable," he says. "It's like a dinosaur." Rossmann, who has also modeled other viruses, like the mengo virus, has gone on to produce the image of the site where an antiviral drug binds to the surface of a virus -- important in both understanding how existing drugs work and developing new ones.

Anton Hopfinger, a chemist at the Chicago campus of the University of Illinois, is using computer graphics to identify the site where adriamycin, a chemotherapy drug, binds to cancer cells. "Molecular graphics has been a real boon to the study of large molecules and proteins," he says. "You can think of it as the equivalent of landing an airplane on an aircraft carrier, except in this case you're sitting on the drug molecule and landing on the DNA molecule. If you didn't have graphics, it would be like being blind and still trying to land on the aircraft carrier."

Some scientists warn against going overboard with the new technique. Says James Blinn, a Jet Propulsion Laboratory scientist who created some of NASA's most spectacular computer simulations of planetary flybys: "Sometimes a half- baked idea gets printed up prettily and gets more attention than it deserves." Still, Blinn believes, as long as the scientific data used to generate the images are accurate, computer graphics can prod scientists to move in exciting new directions. NCSA's Upson agrees. "If we play our cards right," he says, "we may actually make a dent in how people do science."

With reporting by Cristina Garcia/San Francisco and J. Madeleine Nash/Chicago